No face recognition in new images

[SysTechIL]

I have searched the existing issues, both open and closed, to make sure this is not a duplicate report.

• Yes

The bug

When uploading new images, the AntelopeV2 face recognition model does not detect any faces at all.
I expected the faces in the images to be recognized automatically.
Switching to the Buffalo_L model detects most faces correctly.
Even after running the Re-extract process, many faces still do not appear in the interface.

The OS that Immich Server is running on

VM Ubuntu 22.10 on HyperV

Version of Immich Server

1.140.1

Version of Immich Mobile App

1.140.1 build 220

Platform with the issue

• Server
• Web
• Mobile

Device make and model

No response

Your docker-compose.yml content

services:
  immich-server:
    container_name: immich_server
    image: ghcr.io/immich-app/immich-server:${IMMICH_VERSION:-release}
    volumes:
      - immich_upload:/usr/src/app/upload
      - /mnt/immich-photos:/mnt/immich-photos
    env_file:
      - ./immich.env
    ports:
      - '2283:2283'
    depends_on:
      - redis
      - database
    restart: always
    healthcheck:
      disable: false

  immich-machine-learning:
    container_name: immich_machine_learning
    image: ghcr.io/immich-app/immich-machine-learning:${IMMICH_VERSION:-release}
    volumes:
      - model-cache:/cache
    env_file:
      - ./immich.env
    restart: always
    healthcheck:
      disable: false
  redis:
    container_name: immich_redis
    image: docker.io/valkey/valkey:8-bookworm@sha256:00000000
    healthcheck:
      test: redis-cli ping || exit 1
    restart: always

  database:
    container_name: immich_postgres
    image: ghcr.io/immich-app/postgres:14-vectorchord0.4.3-pgvectors0.2.0@sha256:000000000
    env_file:
      - ./immich.env
    environment:
      POSTGRES_PASSWORD: ${DB_PASSWORD}
      POSTGRES_USER: ${DB_USERNAME}
      POSTGRES_DB: ${DB_DATABASE_NAME}
      POSTGRES_INITDB_ARGS: '--data-checksums'
    volumes:
      - immich_postgres_data:/var/lib/postgresql/data
    shm_size: 128mb
    restart: always

volumes:
  immich_upload:
  immich_postgres_data:

Your .env content

DB_USERNAME=immich
DB_PASSWORD=Password
DB_HOST=immich_postgres
DB_PORT=5432
DB_DATABASE_NAME=immich
GEMINI_API_KEY=00000000000

Reproduction steps

1. Upload new images to Immich.
2. Select the AntelopeV2 face recognition model.
3. Run the Re-extract process.
4. Observe that no faces are detected in the interface.

Relevant log output

Additional information

No response

[chrissi5120]

did you reset faces? The models logic is incompatible, i think.. i once tried to switch to antelope but failed for the same reason.

[davidwaze]

Quite same behavior also antelopev2.
When uploading new pictures, don't know why but face detection seems not to be automatically processed.
I need to click manually to the 'missing' button in face detection to get new faces processed.

Didn't pay attention at which version is stopped working bu it worked well before or course.

[Powerbless]

Have also Problems. Get all 2 minutes Errors wirth "failed for all URLs" but the machine learning Container is reachable and Up. Also in the logs that the ML ist become unhealthy and Seconds Later it become healthy back. This toggle all the day over and over...

[alextran1502]

@Powerbless check your machine learning URL settings

[Powerbless]

I checked it and Double checked it. Running "docker exec -it immich-server curl http://immich-machine-learning:3003" in the VM where immich is running i get a good response @alextran1502

`docker exec -it immich_server curl -v http://immich-machine-learning:3003

Host immich-machine-learning:3003 was resolved.

IPv6: (none)

IPv4: 172.19.0.3

Trying 172.19.0.3:3003...

Connected to immich-machine-learning (172.19.0.3) port 3003

using HTTP/1.x

GET / HTTP/1.1

Host: immich-machine-learning:3003

User-Agent: curl/8.14.1

Accept: /

Request completely sent off

< HTTP/1.1 200 OK

< date: Tue, 28 Oct 2025 10:40:31 GMT

< server: uvicorn

< content-length: 23

< content-type: application/json

<

Connection #0 to host immich-machine-learning left intact

{"message":"Immich ML"}`

[SysTechIL]

Just an update - I managed to get it working 🙂

I adjusted the values as follows:

Maximum recognition distance 0.3
Minimum recognized faces 1 - to ensure as many faces as possible are detected
Minimum detection score 0.5 (was 0.6)
Going lower than that (e.g. 0.4) results in inaccurate detections most of the time.

After that, I ran the following jobs:
Face detection - Refresh
Facial recognition - Missing

[proud-nerd]

Quite same behavior also antelopev2. When uploading new pictures, don't know why but face detection seems not to be automatically processed. I need to click manually to the 'missing' button in face detection to get new faces processed.

Didn't pay attention at which version is stopped working bu it worked well before or course.

@davidwaze

I had exactly the same issue as you and in my case it was due to RAM constrains I had in my docker-compose.yml file. Apparently something regarding RAM usage changed and 2GB wasn't enough anymore if all of those automatic task start running at the same time on image upload. You can check with

docker compose logs immich-machine-learning

looking for something like

Worker (pid:14) was sent SIGKILL! Perhaps out of memory?